It is commonly observed in a number of languages that the boundary between adjectives and nouns is not clear cut. This observation is based on a number of syntactic, semantic, and morphological phenomena. 1 This paper attempts to complement these claims by a morphophonological observation, the alternations of presuffixal vowels witnessed in Hungarian. The variable behaviour of certain suffixed nouns and adjectives neatly illustrates the uncertainty about their category. 2
The differences between nouns and adjectives are vague in many respects. This vagueness can be manifested in the meaning of the words concerned and/or their syntactic and morphological distribution. Both adjectives and nouns may be suffixed by the same set of inflectional suffixes, both may occur in the same types of syntactic environments, though not with the same likelihood. We do not intend to give any syntactic or semantic analysis of these environments, we use the minimal syntactic distributional patterns merely to identify the locations where different inflectional suffixes can occur in Hungarian. The syntactic environments verbs occupy are unique and make verbs easily identifiable. This is not the case with adjectives and nouns. 3 By no means do we try to imply that there are no instances of words which unambiguously belong to one of the two categories of nouns and adjectives, but there certainly exist words which cannot be sharply categorized. 4
Our aim in this paper is to show that, at least in Hungarian, the morphological “categories” noun and adjective are but the two extremities of a noun-to-adjective (or adjective-to-noun) scale. A number of both lexemes and word forms seem to be located within the two endpoints of this scale, and can thus be identified only as “more nounlike, less adjectivelike” or “less nounlike, more adjectivelike”. It is easier to examine the formal properties of word forms, their distribution, their phonological shape, than the semantic properties of lexemes, 5 therefore we will pay less attention to the position of lexemes on the noun-to-adjective scale.
We first introduce the relevant characteristics of suffixation in Hungarian, showing that neither the presence or absence of a vowel between stem and suffix (the so-called linking vowel), nor its quality (mid or low due to “lowering”) is motivated only by the phonological shape of these morphs (§1). We then argue that lowering may not only be triggered by stems, but also by suffixes (§2). The next two sections survey the syntactic positions adjectives and nouns can occupy in sentences (§3) and the typical suffixes occurring in these syntactic positions (§4). We then examine how the adjectivalness of suffixes (§5) and stems (§6) affects the quality of the linking vowel. Vowel-final stems are discussed in §7 and a brief section is devoted to the question of whether there is a productive pattern for linking vowels after adjectival stems (§8). Some further examples of how the syntactic position of a word affects its linking vowel are treated in §9. Our findings are summarized in §10.
We are grateful to Ádám Nádasdy and the anonymous reviewers for their comments. Our work is partly sponsored by NKFI grant #119863.
Elekfi, László . 2000. Semantic differences of suffixal alternates in Hungarian. Acta Linguistica Hungarica 47. 145–177.
Halácsy, Péter , András Kornai and Csaba Oravecz . 2007. HunPos: an open source trigram tagger. Proceedings of the 45th Annual Meeting of the ACL [Association for Computational Linguistics]. 209–212.
Hollmann, Willem B. 2021. The ‘nouniness’ of attributive adjectives and ‘verbiness’ of predicative adjectives: Evidence from phonology. English Language & Linguistics 25(2). 257–279.
Kálmán, László , Péter Rebrus and Miklós Törkenczy . 2012. Possible and impossible variation. In F. Kiefer , M. Ladányi and P. Siptár (eds.) Current issues in morphological theory: (Ir)regularity, analogy and frequency. Selected papers from the 14th International Morphology Meeting. Philadelphia, PA: John Benjamins. 23–49.
- Search Google Scholar
- Export Citation
Kálmán, László Péter Rebrus Miklós Törkenczy 2012. Possible and impossible variation. In F. Kiefer M. Ladányi P. Siptár Philadelphia, PA: John Benjamins. 23– 49. 10.1075/cilt.322.02kal
Kiefer, Ferenc (ed.). 2000. Strukturális magyar nyelvtan 3: Morfológia [A structural grammar of Hungarian 3: Morphology]. Budapest: Akadémiai Kiadó.
Kiefer, Ferenc and Mária Ladányi . 2000. Morfoszintaktikailag semleges képzések [Morphosyntactically neutral derivations]. In Kiefer (2000, 165–214).
Lévai, Dániel . 2020. Corpus-based analysis of the variability of linking vowels in nouns and adjectives. MA thesis. Eötvös Loránd University, Budapest.
Moravcsik, Edith . 2001. On the nouniness of Hungarian adjectives. In C. Schaner-Wolles , J. Rennison and F. Neubarth (eds.) Naturally! Linguistic studies in honour of Wolfgang Ulrich Dressler presented on the occasion of his 60th birthday. Torino: Rosenberg & Sellier. 337–346.
- Search Google Scholar
- Export Citation
Moravcsik, Edith 2001. On the nouniness of Hungarian adjectives. In C. Schaner-Wolles J. Rennison F. Neubarth Torino: Rosenberg & Sellier. 337– 346.
Nádasdy, Ádám . 2019. A magyar melléknevek többes száma [The plural of Hungarian adjectives]. Általános Nyelvészeti Tanulmányok 31. 167–185.
Rebrus, Péter . 2019. A magyar névszói inflexiós rendszer a szó és paradigma modellben [The nominal inflection system of Hungarian in the word and paradigm model]. Nyelvtudományi Közlemények 115. 109–147.
Rebrus, Péter and Krisztina Polgárdi . 1997. Two default vowels in Hungarian? In G. Booij and J. van de Weijer (eds.) Phonology in progress – Progress in phonology (HIL Phonology Papers III). Hague: Holland Academic Graphics. 257–275.
Tompa, József . 1957. A névszói kötőhangzó megkülönböztető szerepe [The distinctive role of the nominal linking vowel]. Nyelvtudományi Értekezések 14. Budapest: Akadémiai Kiadó.
This includes the proposed distinction between relational and qualifying adjectives, their order, and selectional restrictions on nominal modifiers. We thank one of our reviewers for pointing this out.
A similar difference in the phonological properties of adjectives occurring in different syntactic positions is noticed by Hollmann (2021) in English.
The fuzziness of categorization is very common. For example, while some objects are clearly green, others are clearly blue, a significant set cannot be obviously assigned to either colour category (or to a third one), but is in the transition zone between the two.
It is not easy to decide whether two occurrences of the same word belong to one or distinct lexemes. This is true for those cases where the two lexemes belong to different categories. Note that in Hungarian offline and online databases a high ratio of adjectives are also tagged as nouns, based partly on theoretical grounds, partly on a lexicographical tradition and on practical considerations.
Since our only concern in this paper is the identity of the vowel between stem and suffix, we use the standard orthographic form of words. The digraph sz represents [s], s is [ʃ], digraphs ending in y are palatal (eg gy is [ɟ]), acute and double acute accent marks on vowel letters and doubling of consonant letters indicate length, ö, ő, ü, ű are front rounded vowels. We insert a hyphen between the stem, the suffix, and, later, the linking vowel, which is not part of the standard orthography.
In glosses we apply the Leipzig glossing rules (https://www.eva.mpg.de/lingua/resources/glossing-rules.php) and Wikipedia's abbreviations (https://en.wikipedia.org/wiki/List_of_glossing_abbreviations) as of June 2021.
Vowel–zero alternation also occurs within stems, either in the root: maj o m ‘monkey’, majm-a ‘monkey-3sg.poss’, or in a suffix: véd-elem ‘protect-nmz’, véd-elm-e ‘protect-nmz-3sg.poss’, and at the end of a free stem: barn a ‘brown', barn-ít ‘brown-vbz'. We are not concerned with these alternations in the present paper.
The vowel inventory of Budapest Hungarian contains seven short vowels, i, ü, u, e, ö, o, and a.
Note that the 3pl possessive suffix -u/ük (e.g., dal-uk ‘song-3pl.poss’, fül-ük ‘ear-3pl.poss’) does not seem to be a linking vowel, with singular nouns it does not alternate with zero: kocsi-j-uk ‘car-3pl.poss’, olló-j-uk ‘scissors-3pl.poss’. If anything, the -j- is a linking consonant here. However, we do see alternation in the case of plural nouns: dal-u-k ‘song-3pl.poss’, dal-ai-k ‘song-pl-3pl.poss’. We leave this issue open.
See, however, Rebrus (2019) for an alternative proposal that does not require morphological segmentation.
The linking vowel is present after some word-final consonant clusters (e.g., fals-o-t ‘fake-acc’) and absent after others (e.g., docens-t ‘associate professor-acc’). We do not go into details here, see Kálmán et al. (2012).
Palatal consonants count as coronal, so typically there is no mid linking vowel in accusative forms after j or ny (e.g., baj-t ‘trouble-acc', lány-t ‘girl-acc').
Some accents of Hungarian distinguish between two nonhigh front unrounded vowels, a mid and a low one. In such an accent, the front unrounded linking vowel will be low(er) after a lowering stem than after a nonlowering stem. In the Budapest accent these vowels are merged and the orthography does not distinguish them either.
The distinction between inflectional and derivational suffixes is not always obvious and often theory specific. We ignore the details here.
The possessive forms of adjectives are rather infrequent and usually missing from corpora, so we had to rely on our intuitions in this case.
We find -öbb in the suppletive form több ‘more’ (cf. sok ‘many’), where the vowel is etymologically part of the stem. Unexpectedly -o- occurs with comparative -bb in a single adjective, nagy ‘big’: nagy-o-bb ‘bigger’. The -o- in jobb ‘better’ is also part of the stem jó ‘good’. Note that free stems never end in short -o or -ö.
In verbs the presence vs. absence of a linking vowel is not a manifestation of lowering. Typically we find a linking vowel after clusters, and not after single consonants. Thus, although we find lowering before the infinitive and the conditional suffixes, there is no linking vowel in öl-ni ‘kill-inf’ or lát-na ‘see-cond’, but there is one in ölt- e -ni ‘stitch-inf’ or bánt- a -na ‘hurt-cond’. For details see Siptár & Törkenczy (2000), Rebrus (2000).
In this paper, we use the labels in (6) uniformly for nouns and adjectives in a distributional sense without siding with any particular syntactic analysis. Attributive position means ‘prenominal modifier’, predicative position means ‘copular complement’, NP-head means an overt noun in the head position or an adjective modifying the ellipted noun head of an NP. This categorization is motivated by the uniform distribution of inflectional suffixes in each type, see (8).
Note how English uses one and -ing to adopt the adjective in a nominal and the noun in an adjectival position, respectively. Also note that neither a noun, nor an adjective requires a determiner in predicative position in Hungarian, while the noun does, but the adjective does not in English.
There are specific syntactic constructions containing noun+noun which are not rare, e.g., where the first noun denotes a profession: orvos barát-om ‘lit. doctor friend-1sg.poss (my friend who is a doctor)', tanár feleség ‘lit. teacher wife (a wife who is a teacher)', nyelvész lány ‘lit. linguist girl (a girl who is a linguist)'.
A determiner+adjective construction is analysed in many syntactic theories as a phrase containing an ellipted nominal head which is modified by the adjective. Morphologically, however, this is irrelevant: it is the adjective that is suffixed in this construction.
The word nominal may mean ‘sharing features with both nouns and adjectives’. Here we only use it as the adjectival form of the noun noun, meaning ‘nounlike’.
Many nouns do not occur in attributive position, but have to be suffixed with an adjectivizer, e.g., *város ház ‘city house’ vs. város-i ház ‘city-adjz house’; while both the noun szomszéd and the adjective szomszéd-os ‘neighbour-adjz’ can appear in this position: szomszéd(-os) ház.
Even this morpheme seems to have a lexically conditioned allomorph, -ul/ül (konok-ul ‘stubborn-ly’, pazar-ul ‘sumptuous-ly’, szó-tlan-ul ‘word-less-ly’, német-ül ‘in German’, etc.). Intriguingly, this allomorph is homonymous with the so-called essive-modal suffix productively added to nouns (mintá-ul ‘as a pattern’, emlék-ül ‘as a memory’). This also hints at a lack of a clear-cut distinction between denominal and deadjectival suffixes.
Note, however, that with some vowel-final adjectival stems these suffixes take an s-initial allomorph, which, in fact, is the productive pattern (e.g., forró-sít ‘hot-vbz’, hiú-sul ‘vain-vbz’, sűrű-södik ‘dense-vbz’). The same strings can productively be added to nominal stems, whether they be analysed an adjectivizer -s followed by the verbalizer or as a single morpheme (e.g., tanú-sít ‘witness-vbz’, íz-esül ‘joint-vbz’, szaru-sodik ‘callus-vbz’). We again witness an overlap between adjectives and nouns as the base of suffixation.
English seems to apply a different, perhaps phonological constraint here: adjectives ending in -(i/e)an may be used as a “noun”, others, for example those ending in [ʃ] or [tʃ], may not: an American/Athenian/Ethiopian vs. *a Welsh/French/Polish.
For example, while adjectives formed by the adjectivizer -s generally lower, they often do not have a linking vowel (i.e., do not lower) before the accusative suffix: melyiket kéred? a két-ablak-o-s-( a -)t ‘which one do you want? the one with two windows (lit. the two-window-adjz-acc)’. This may be due to the large number of lexicalized nouns ending in the same suffix: e.g., asztal-o-s ‘lit. table-adjz (=carpenter)’. The privative suffix -t(a)lan/t(e)len forms adjectives, which all belong to the nonlowering “irregular adjective” group.
Piréz is an imaginary ethnic group name made up by a Hungarian polling company in 2006 in a survey of xenophobia: https://en.wikipedia.org/wiki/Piréz_people.
Being a front unrounded stem, piréz shows the lack of lowering only with the accusative suffix: piréz-t. The ad-hoc creation varéz exhibits the same lack of lowering with other suffixes too: the plural is varéz- o -k, the 2sg possessive is varéz- o -d, etc.
Since we here have words with both back and front harmony, we use the standard variables “A” for -a- and -e-, the low linking vowel, and “O” for -o-, -ö-, and -e-, the mid linking vowel.
Word-final long “low” vowels occur rarely and remain unchanged: ordenáré ‘gross’, ordenáré-k ‘gross-pl’.
László Fejes calls our attention to the marginal plural forms including lowering in gané ‘scummy’, gané-(?a)k ‘scummy-pl’; csálé ‘bevelled’, csálé-(?a)k ‘bevelled-pl’.